Also in few cases see inconsistency where IAM throws 429 but still processes the request thus creating a state drift.
A way to handle this might be to do a GET check on resource in case of error before retrying request.
Need to validate further, but i guess inconsistency was caused by multiple retries. If after rate limit error leave the system for cool-off seems eventual consistency plays & resources are deleted.
Provider should be extended to do backoff retry for rate limited resources.
Example:
│ Error: DELETE /authorize/scim/v2/Applications/fc527f11-900f-48f9-a81b-038e8cb96cae: StatusCode 429, Body: {"issue":[{"details":{"coding":{"userSelected":false},"text":"Rate limit exceeded"},"diagnostics":"Rate limit exceeded"}]}
Also in few cases see inconsistency where IAM throws
429
but still processes the request thus creating astate drift
. A way to handle this might be to do a GET check on resource in case of error before retrying request.